October 8, 2024:
In this article we’re going to talk about market timing and why it’s so hard to get right, including some entertaining evidence from a new study. After that we’ll do a quick tour through some interesting items regarding stock market opportunities, AI, and renting vs. buying in San Diego.
Thinking about market timing?
Clients sometimes ask whether they should change their investment position based on something happening in the news, such as an election, a change to Fed policy, or a geopolitical event.
We understand that it can be tempting to try to time market exposure based on headline events. Sometimes, it really just feels so obvious that this or that event will lead to a given market outcome. But the reality is that market timing is very difficult to get right, very unreliable, and (we think) more likely to hurt than help.
Here are some of the reasons why:
- It’s not enough to be right. Market prices represent the collective views of all market participants, which is a fancy way of saying, “If people know about it, it’s already priced in.” This means that it’s not enough for you to correctly forecast an event: for the market to respond the way you want it to, you have to have a different forecast than most other market participants, and you have to be more right than them.
- The markets are too complex to boil down to one factor. Even if you pull off the difficult task of forecasting an event better than most, you could still end up losing money. Markets, like the world economy that drives them, are incredibly complex. You could nail the specific event you predicted, only to have the market go against you as it gets overwhelmed by some of the million other forces that are impacting markets at any given time.
- The first decision is only half the battle. It’s not enough to make the initial trade; you also have to figure out when to reverse it — which is just as unreliable and maybe even more so. If you don’t get that second decision right, it’s easy to get trapped in a bad and very stressful situation.
- It’s fertile ground for emotional decision-making. Making “calls” based on headline events is highly subjective, and so vulnerable to being influenced by feelings or cognitive biases. This risk grows worse if there are unusually scary headlines or big moves in the markets.
A recently-published paper describes an experiment run by the authors: they chose 15 days from the past, taken randomly from each year running from 2008 to 2022. They let a group of finance-trained investors place whatever trades they wanted for those days. Here’s the twist: for each day, investors were given a copy of the next day’s Wall Street Journal front page (with stock and bond results removed) before they made their trades.
You’d think seeing the next day’s news would be a huge advantage. But on average, the traders’ results were indistinguishable from flipping a coin. And 1 out of every 6 of them managed to go bust! Investors are bad at market timing even if they have a crystal ball.
In fairness, the authors also tried this with a small group of veteran professional macro traders, and they did a lot better. But the fact that the larger group of financially savvy traders couldn’t make money even when they could see into the future shows just how hard market timing can be.
Quick hits!
Here are some interesting things we’ve seen lately. The first two concern opportunities in areas of the stock markets that we are emphasizing in our portfolios (see our previous article for a bit more detail).
Growth vs. value stocks
Value stocks outperformed growth stocks last quarter, but it looks like there is a lot of room for that trend to continue. This chart shows the relative valuation between growth and value stocks (when the line is low, value stocks are comparatively inexpensive). The chart ends in June, so it doesn’t show last quarter’s value outperformance. But that won’t have changed the big picture: there is a long way to go before this relationship is back in line.
US vs. international stocks
International stocks (mildly) outperformed US stocks last quarter, and we think there’s a lot of room to run there as well.
The chart below shows different stock regions’ price/forward earnings ratios (a valuation measure that compares stock prices with the next year’s forecasted company earnings). The red diamond is the current forward p/e, and the darker blue square is the 20 year median. The light blue bar covers the middle 50% of values over the past 20 years.
In contrast, non-US stocks (the four rightmost bars) are all pretty close to their 20-year medians and within their normal ranges. And they are all substantially less expensive than US stocks, whether big tech is included or not.
Artificial Intelligence
Generative AI can do some really incredible things. But GenAI models are very expensive to build and train, and there is a lot of debate about whether their actual economic benefits will be sufficient to justify all that spending. We enjoyed this Goldman Sachs piece featuring a wide spectrum of viewpoints on the Generative AI cost-benefit question.
Along those lines, here’s a good snippet from financial writer Brent Donnelly. It uses a bit of jargon: LLM = “large language model” (a type of GenAI such as the one that powers ChatGPT); AGI = “artificial general intelligence,” which implies human-like intelligence across a wide range of tasks. Brent speculates why many AI-related stocks were struggling for much of this year:
More and more people I talk to and read are coming to the conclusion that:
- LLM does not lead to AGI or even smartish AI.
- LLMs do not obey Moore’s Law [the rough doubling of computer capacity every two years] because you can’t jam an infinite number of Reddit posts and copyrighted material into them forever. There is an upper bound on the content available for processing.
- LLMs are a commodity and every company in the world is creating one and marketing it as the next big thing, but in the end it’s gonna be hard to monetize a commodity that’s essentially undifferentiated and free.
And a final quick quote: computer programmer Simon Willison described training LLMs as “money laundering for copyrighted data.” Tough but fair! We have yet to see how courts and regulators will approach the intellectual property issue.
There are a lot of potential economic benefits from Generative AI, but a lot of open questions too. Only time will tell how well GenAI’s future matches up with what’s already been priced into markets.
San Diego housing affordability
Switching gears to our local housing market, here are a couple of charts illustrating the cost of buying vs. renting a home in San Diego. The first compares home prices to rents, and the second compares monthly home payments to rents. Both measures show buying to be unusually expensive compared to renting right now.
